A 5-year national linked primary and secondary care cohort study


Methods and findings

We constructed a national cohort, identified from 76% (2.4 million) of the Welsh population, of continuously treated asthma patients between 2013 and 2017 using anonymised, person-level, linked, routinely collected primary and secondary care data in the Secure Anonymised Information Linkage (SAIL) Databank. We investigated the association between asthma-related health service utilisation, prescribing, and deaths with the 2011 Welsh Index of Multiple Deprivation (WIMD) and its domains. We studied 106,926 patients (534,630 person-years), 56.3% were female, with mean age of 47.5 years (SD = 20.3). Compared to the least deprived patients, the most deprived patients had slightly fewer total asthma-related primary care consultations per patient (incidence rate ratio [IRR] = 0.98, 95% CI 0.97–0.99, p-value < 0.001), slightly fewer routine asthma reviews (IRR = 0.98, 0.97–0.99, p-value < 0.001), lower controller-to-total asthma medication ratios (AMRs; 0.50 versus 0.56, p-value < 0.001), more asthma-related accident and emergency (A&E) attendances (IRR = 1.27, 1.10–1.46, p-value = 0.001), more asthma emergency admissions (IRR = 1.56, 1.39–1.76, p-value < 0.001), longer asthma-related hospital stay (IRR = 1.64, 1.39–1.94, p-value < 0.001), and were at higher risk of asthma-related death (risk ratio of deaths with any mention of asthma 1.56, 1.18–2.07, p-value = 0.002). Study limitations include the deprivation index being area based and the potential for residual confounders and mediators.

Author summary


Asthma is one of the most prevalent chronic diseases and has significant clinical and economic burden [1]. However, asthma burden is not evenly distributed within populations, and socioeconomic variations in asthma prevalence, emergency hospital admissions, and mortality have been recorded worldwide [24]. These variations have been attributed to a range of modifiable and non-modifiable factors. Ethnicity has been found to partly explain the higher asthma prevalence, disease severity, and risk of asthma admissions among South Asians and Afro-Caribbeans in the United Kingdom [5,6] and African-Americans and Puerto Ricans in the United States [7]. However, lower household income has been identified as an independent risk factor for the development of persistent asthma among children [8] and for worse asthma outcomes [9]. Suboptimal asthma self-management and worse asthma outcomes have been associated with lower health literacy [1012]. Air quality has also been linked to asthma severity [13,14], although the literature on its association with asthma incidence and prevalence is inconclusive [15,16].

The UK has a high asthma prevalence and burden [1] and persistent socioeconomic inequalities where the more deprived people had worse asthma outcomes including higher risk of emergency admissions and deaths due to asthma [3,17]. In Wales, previous studies found that severe asthma admissions were more likely in the most deprived areas [18,19], which was attributed to active or passive smoking, variations in disease management, and air pollution. However, asthma is mostly managed in primary care in the UK with hospital care reserved for those with more severe disease or, most commonly, in the context of asthma attacks. Therefore, in order to comprehensively investigate inequalities in asthma care in this country, there is a need to study asthma care provision across care sectors. There is currently sparse evidence on the socioeconomic inequalities in asthma-related primary care, prescribing, accident and emergency (A&E) use, and deaths across the different demographic groups.

In this study, we investigated socioeconomic variations in asthma care and outcomes in Wales, including asthma-related primary and secondary health service utilisation, prescribing, and mortality, using a multifaceted measure of socioeconomic deprivation as well as individual domains of deprivation.


Study design and data sources

We undertook a national, linked primary and secondary care retrospective cohort study of people with asthma in Wales. The follow-up period was 5 years from January 1, 2013 to December 31, 2017.

We used anonymised linked person-level datasets about primary and secondary care and causes of death from the Wales-wide SAIL Databank [20,21]. SAIL has 100% coverage in Wales for secondary care and causes of death and receives data from over 76% of the Welsh general practices. S1 Text provides more details about data sources and study design, and S2 Text details criteria of patient selection. S1 Table lists the code sets used in the patient selection and extraction of variables.


Socioeconomic status.

We measured socioeconomic status using the 2011 version of the Welsh Index of Multiple Deprivation (WIMD), the official area-based measure of relative socioeconomic deprivation in Wales [22]. The WIMD 2011 was constructed from a weighted sum of 8 deprivation domains: income (23.5%), employment (23.5%), health (14.0%), education (14.0%), geographical access to services (10.0%), housing (5.0%), physical environment (5.0%), and community safety (5.0%). The WIMD 2011 is based on Lower Layer Super Output Areas (LSOAs) which is a small area geography designed by the UK Office for National Statistics (ONS) for census-related purposes with consistent population sizes (1,500 people on average) [22]. We linked the patients to the WIMD through their residential addresses (LSOAs of the 2001 Census [23]) during the follow-up period. Where more than 1 address existed, we selected the address with the longest duration within that period. Quintiles of score were coded from 1 for the most deprived to 5 for the least deprived. S1 Fig shows the distribution of the WIMD 2011 score and its quintiles.

Asthma-related health service utilisation.

Asthma-related health service utilisation was measured as counts of the corresponding primary and secondary care events, length of stay (LOS) in hospital, and controller-to-total asthma medication ratio (AMR) during the follow-up period.

We defined an “asthma-related general practitioner (GP) consultation” as 1 or more Read codes that indicated asthma-related contact with primary care professionals.

An asthma review was defined as scheduled consultations to a primary care practice in which disease control is assessed, and management plan, prescriptions, and asthma self-management advice were reviewed. The British Thoracic Society/the Scottish Intercollegiate Guidelines Network (BTS/SIGN) Guidelines on the Management of Asthma recommends that asthma reviews be arranged at least annually [24]. We identified routine asthma reviews from the Welsh Longitudinal General Practice (WLGP) dataset using codes of annual review, medication review, follow-up, monitoring by nurse, and review using the Royal College of Physicians’ 3 Questions for Asthma [25].

We identified asthma-related A&E attendances with primary or secondary diagnosis of asthma from the Emergency Department Data Set using the dataset-specific asthma code 14A.

We identified asthma admissions from the Patient Episode Database for Wales as those with a primary diagnosis of asthma (J45) or status asthmaticus (J46) coded using the 10th Revision of the International Classification of Diseases (ICD-10). Among these, emergency admissions were defined as coming via A&E departments, urgent referrals from GPs, consultant clinics, bed bureaus, or NHS Direct [26].

Asthma medication ratio.

The AMR, the ratio of controller-to-total asthma prescriptions, has been developed in the US as a surrogate quality measure of guideline adherence and is associated with patient outcomes and health service utilisation [27]. AMR calculation included counts of inhaled corticosteroids (ICS), ICS-long-acting beta adrenoceptor agonist (LABA) combination inhalers, sodium cromoglicate, and nedocromil as controller prescriptions, and short-acting beta agonist (SABA) inhalers as rescue prescriptions over the follow-up period. The formula was (ICS + ICS_LABA + sodium cromoglicate + nedocromil)/(ICS + ICS_LABA + sodium cromoglicate + nedocromil + SABA).

Statistical analysis

For the source population, we described demographics, point prevalence of ever being diagnosed with asthma (i.e., having an asthma diagnosis Read code) on January 1, 2013, and the period prevalence of ever-diagnosed currently treated asthma during 2013 (having an asthma diagnosis Read code ever and at least 1 asthma prescription code during 2013).

For the study cohort, we calculated the distribution, by the WIMD quintile, of age, gender, receipt of asthma prescription categories, and the health service utilisation variables. For AMR, we excluded patients who did not receive any of the prescriptions in the formula.

For each health service utilisation count variable, we fitted a negative binomial (NB) generalised linear model using the glm.nb function from the MASS package (version 7.3–51.4) [28]. We considered the least deprived quintile (WIMD 5) as the reference group. We adjusted the models for gender and age at the start of follow-up. We treated LOS in hospital as a count variable (count of days) for which the model coefficient represents the incidence rate ratio (IRR) of incurring an additional day in hospital in a given quintile compared with the least deprived quintile. Model fit was examined using quantile–quantile (Q–Q) plots of raw residuals and rootograms [29]. These models were specified a priori. However, we performed 2 separate sensitivity analyses; in the first one, we removed the condition of having continuous asthma treatment over the follow-up period, and in the second one, we removed the condition of continuous follow-up in the primary care dataset. We also modelled these count outcome variables using generalised additive models (GAMs) using the R package mgcv (version 1.8–31) [30]. We initially estimated global smooths for both the overall WIMD score and age. Then, we explored the within-gender effect of the overall WIMD score by estimating separate smooths for males and females. From these, we calculated a difference smooth [31] to explore the between-gender variation in the effect of the overall WIMD score. We then modelled the interaction between the overall WIMD score and age using a full tensor product smooth (mgcv::te). The count variables were also modelled separately against the score of each WIMD domains, controlled for age and gender.

We compared the mean AMR between the most and least deprived quintiles using Welch t test. Then, we fitted a GAM using the beta regression family (mgcv::betar), adjusted for age and gender, to explore the associations with the overall WIMD score.

We modelled each definition of asthma deaths using logistic regression of WIMD quintile adjusted for gender and age. The effect of gender was then examined in separate models within each WIMD quintile. We then fitted binomial GAM models, including global smooths for the overall WIMD score and age, smooths by gender, and a difference smooth by gender, and compared the overall fitted risk of asthma death between males and females across the overall WIMD score. Finally, asthma deaths were then modelled separately against the score of each WIMD domains, controlled for age and gender.

All the GAMs in this paper used a thin plate regression spline as a smoothing basis and the fast restricted maximum likelihood (fREML) computation as a smoothing parameter estimation method.

We compared the ratio of overall emergency-to-total hospitalisations for asthma between the WIMD quintiles using equality of proportions test.

We used a confidence level of 95% (p < 0.05, 2-sided) throughout the study. All data analysis was performed in R 4.0.2.

Study planning

Investigating asthma inequalities in Wales was a planned part of the first author’s doctoral thesis about creating and utilising the Wales Asthma Observatory. There was no prospective protocol for the study. Health service utilisation count regression models against the WIMD 2011 score quintile were specified a priori, and preliminary results for an earlier follow-up period were previously published [35,36]. To provide a more comprehensive picture about inequalities, the analysis was developed to include wider outcomes such as asthma emergency admissions, prescribing, and deaths as well as the WIMD 2011’s individual domains as predictors. The GAMs were developed in response to a peer review request to investigate nonlinear associations.


Table 1 shows characteristics of the source population (n = 2,871,257). The prevalence of ever-diagnosed asthma in the source population at the beginning of 2013 was 11.9% (95% CI, 11.8 to 11.9), with 12.5% (12.5 to 12.6) in the most deprived areas and 11.7% (11.6 to 11.8) in the least deprived areas. The prevalence of ever-diagnosed currently treated asthma during 2013 was 7.2% (7.1 to 7.2), ranging from 7.7% (7.6 to 7.8) to 6.8% (6.7 to 6.9) in the most and least deprived areas, respectively.

The study cohort included 106,926 patients (534,630 person-years) with ever-diagnosed, continuously treated asthma. Patient selection flowchart is shown in S2 Text, and Table 2 shows the patients’ characteristics. Females comprised 56.3% of patients (58.9% to 54.2% in the most and least deprived areas). Mean age was 47.5 years (SD = 20.3).

Patients in the most deprived quintile (WIMD 1) represented the highest proportion (23.4%) of the study cohort and were younger overall (mean age = 45.6 years, SD = 20.1) than those in the least deprived areas (48.8 years, SD = 20.4). At least 1 asthma-related GP consultation and 1 asthma review was recorded over the follow-up period for 98.5% and 95.0% of patients, respectively. Only 2.4% and 3.6% had asthma-related A&E attendances and hospitalisations during the follow-up period.

Table 3 and Fig 1 show the estimated associations of the WIMD 2011 quintile with asthma-related primary and secondary care utilisation controlled for age and gender. The count regression models showed good fit to the data (see S2 Fig). The estimates did not significantly change by relaxing the patient selection criteria to also include patients with any asthma treatment status or any follow-up periods in the primary care dataset (see S3 Text).


Fig 1. IRRs with 95% CIs of asthma health service utilisation in each of the WIMD 2011 quintiles relative to the least deprived quintile, controlled for age and gender.

Note: The CIs for GP consultations and review are extremely narrow and may not be obvious in the plot. A&E, accident and emergency; CI, confidence interval; GP, general practitioner; IRR, incidence rate ratio; WIMD, Welsh Index of Multiple Deprivation.


Asthma-related primary care consultations and reviews

The most deprived quintile (WIMD 1) had %1.9 fewer GP consultations (IRR = 0.98 [95% CI, 0.97 to 0.99], p-value < 0.001) and %2.0 fewer routine asthma reviews (IRR = 0.98 [95% CI, 0.97 to 0.99], p-value < 0.001) per patient compared with the least deprived quintile (WIMD 5). The WIMD quintiles 2 to 4 were also associated with slightly fewer asthma reviews than the least deprived quintile. The corresponding GAMs support these association patterns (Fig 2). Less geographical access to services was associated with slightly more asthma GP consultations and slightly fewer asthma reviews (Fig 3). Overall, females had 2.4% more asthma-related GP consultations (IRR = 1.02 [1.02 to 1.03], p-value < 0.001) and 5.0% more asthma reviews (1.05 [1.04 to 1.06], p-value < 0.001) than males. However, these gender gaps had minor variations across the overall WIMD score (Figs 4 and 5) and age: In middle age, females had slightly higher rates than males, whereas males had slightly higher rates among children and the older adults. In both genders, however, there was a minor variation across age, with asthma reviews being lowest in middle age, whereas the younger people had slightly more asthma-related GP consultations (Fig 6).


Fig 2. Smooths from GAMs showing the association of the WIMD 2011 score with asthma-related health service utilisation.

The shaded area represents the 95% CI of effect. A&E, accident and emergency; CI, confidence interval; GAM, generalised additive model; GP, general practitioner; WIMD, Welsh Index of Multiple Deprivation.



Fig 3. Smooths from separate GAMs of asthma-related health service utilisation variables against scores of the WIMD 2011 domains, controlled for age and gender.

The shaded area represents the 95% CI of effect. A&E, accident and emergency; CI, confidence interval; GAM, generalised additive model; GP, general practitioner; WIMD, Welsh Index of Multiple Deprivation.



Fig 4. Difference smooths from GAMs showing gender variations in the effect of the WIMD 2011 score on asthma-related health service utilisation.

The shaded area represents the 95% CI of effect. A positive difference means a higher effect in females than males. The gender gap was stable for asthma primary care consultations and reviews and variable for A&E attendances, emergency admissions, and LOS. A&E, accident and emergency; CI, confidence interval; GAM, generalised additive model; GP, general practitioner; LOS, length of stay; WIMD, Welsh Index of Multiple Deprivation.



Fig 5. Full tensor product smooths from GAMs of asthma-related health service utilisation showing interaction between the WIMD 2011 score and age in males and females.

The colour represents the partial effect (blue = lower, red = higher). The models did not include separate smooths for the WIMD 2011 score and age. A&E, accident and emergency; CI, confidence interval; GAM, generalised additive model; GP, general practitioner; WIMD, Welsh Index of Multiple Deprivation.



Fig 6. Smooths from GAMs showing the variation in asthma-related health service utilisation across age.

The shaded area represents the 95% CI of effect. A&E, accident and emergency; CI, confidence interval; GAM, generalised additive model; GP, general practitioner.


Asthma prescriptions in primary care

Patients in the most deprived areas had more primary care asthma prescriptions, at 17.6 prescriptions per year, compared with 12.0 in the least deprived areas. A socioeconomic gradient existed for all the classes of asthma medications (Table 2). On average, people from the most deprived areas received 7.3 reliever (SABA) inhalers and 7.3 controller prescriptions of ICS, ICS-LABA, sodium cromoglicate, and/or nedocromil per year, whereas those in the least deprived areas received 4.6 relievers and 5.7 controllers. Patients in the most deprived areas were 3.0 times more likely to have 12 or more SABA inhaler per year (risk ratio = 3.0 [2.8, 3.2], p-value < 0.001).

Mean AMR (controller-to-total medication ratio) was lower in the most deprived quintile (0.50) than in the least deprived quintile (0.56; Welch t test p-value < 0.001; ratio of means = 90.3% [89.6% to 91.0%]; absolute difference in means = 0.054 [0.050 to 0.058]). The GAM showed that AMR generally decreased with higher WIMD score up to around the overall WIMD score of 40 before increasing slightly in the higher WIMD scores (Fig 7). However, AMR distribution below 0.5 was similar across the WIMD quintiles, whereas fewer patients had AMR >0.5 in the more deprived quintiles (see the empirical cumulative distribution function plot in S3 Fig). The variation of AMR across age was greater than across the WIMD score, with the late teens and early 20s having the lowest values. There was no statistically significant difference in AMR between males and females (odds ratio: 1.003 [0.989 to 1.017], p-value = 0.721).


Fig 7. A GAM of AMR by the WIMD 2011 score and age, adjusted for gender.

The shaded area represents the 95% CI of effect. AMR, asthma medication ratio; CI, confidence interval; GAM, generalised additive model; WIMD, Welsh Index of Multiple Deprivation.


Asthma-related A&E attendances

The most deprived areas had 26.9% more A&E attendances than the least deprived areas (IRR = 1.27 [1.10 to 1.46], p-value = 0.001). The IRR decreased to 1.23 [1.07, 1.41] after controlling for AMR (p-value = 0.004). No gradient existed across the deprivation quintiles, but there was a contrast between the least deprived quintile and the other more deprived quintiles together, a pattern also seen in the corresponding GAM (Fig 2). Lower education levels and worse housing were consistently associated with higher attendance rates. Higher rates were also seen towards lower levels of income, employment, and general health (Fig 3). However, rates were lower with less geographical access to services and worse physical environment.

Overall, females had 62.4% more asthma-related A&E attendances than males (IRR = 1.62 [1.48 to 1.78], p-value < 0.001). However, the gender gap was variable across age and deprivation; in middle age, females had higher rates than males, whereas males had higher rates in childhood and also among the older adults living in the most deprived areas (Figs 5 and 8). Overall, rates were highest in the youngest patients and steeply decreased in older ages (Figs 5 and 6).


Fig 8. Predicted 5-year counts of health service utilisation events for the median age in males and females based on GAMs of asthma-related health service utilisation by the WIMD 2011 score.

The shaded areas represent the 95% CIs of predicted counts. A&E = general practitioner; GP = general practitioner; WIMD, Welsh Index of Multiple Deprivation.


Asthma hospitalisations

A steep socioeconomic gradient existed for emergency and total asthma-related hospitalisations (Figs 1 and 2). Patients in the most deprived quintile were 55.9% more likely to require emergency admissions for asthma (IRR = 1.56 [1.38 to 1.76], p-value < 0.001) than those in the least deprived quintile. The IRR decreased to 1.52 [1.35 to 1.72] after controlling for AMR (p-value < 0.001). Patients in the most deprived quintile were 97.7% more likely to require any hospitalisation (whether emergency or elective) for asthma (IRR = 1.98 [1.74 to 2.25], p-value < 0.001) than those in the least deprived quintile. Asthma patients in the most deprived areas had 64.0% longer asthma-related hospital stay than those in the least deprived areas (mean of 0.25 versus 0.15 days during the 5-year follow-up period, respectively; IRR = 1.64 [1.39 to 1.94], p-value < 0.001). The GAMs support these patterns, although they show variation within the most deprived quintile with the highest admission rates being around the overall WIMD score of 60 (Fig 2). Higher rates of emergency admissions were associated with worse scores in all the WIMD domains except housing. However, the rates decreased with less geographical access to services (Fig 3). At the WIMD quintile level, however, the most deprived quintile had the lowest proportion of emergency-to-total asthma-related admissions (66.2%), whereas the least deprived quintile had the highest proportion (85.4%, S4 Fig). In the whole cohort, 57.1% of the nonemergency admissions were day cases with no overnight stay.

Overall, females had 66.7% more emergency admissions (1.67 [1.54 to 1.80], p-value < 0.001), 60.6% more total hospitalisations (1.61 [1.48 to 1.75], p-value < 0.001), and 98.2% longer hospital stay related to asthma (1.98 [1.77 to 2.22], p-value < 0.001) than males. However, the gender gap showed some variations across the overall WIMD score and age; females had higher admission rates and longer hospital stay in middle age and also among the most deprived older adults, whereas males had higher rates among children and also among the less deprived older adults (Figs 5 and 8).

Children, especially males, had higher asthma admissions rates and longer stay in hospital than the other age groups (Fig 6).

Asthma-related deaths

In the wider cohort of 327,906 asthma patients, 543 had death with any mention of asthma and 207 had death with asthma as the underlying cause over the study period. Risk of death with either definitions generally increased with higher deprivation (Fig 9). Asthma patients in the most deprived quintile were 56.3% more likely to have death with any mention of asthma within 5 years than those in the least deprived areas (risk ratio, RR = 1.56 [1.18 to 2.07], p-value = 0.002, Table 4). When asthma deaths were identified by the underlying cause only, we could not detect differences between the least deprived and other quintiles, although deaths were associated with the overall WIMD score (Fig 9).


Fig 9. Smooths from GAMs of asthma deaths by the WIMD 2011 score, controlled for age and gender.

The shaded area represents the 95% CI of effect. CI, confidence interval; GAM, generalised additive model; WIMD, Welsh Index of Multiple Deprivation.


Among females, risk of asthma deaths generally increased with deprivation, whereas among males, the highest risk was in the middle WIMD quintile (around the score of 19, Fig 10). Females were generally at higher risk of asthma deaths (Fig 11). However, the gender gap varied across the overall WIMD score and was wider (i.e., females having higher risk) in the most deprived (risk ratios for deaths with asthma as the underlying cause, RR = 4.16 [1.76 to 9.80], p-value = 0.001) and in the least deprived quintiles (2.74 [1.20 to 6.24], p-value = 0.016) and diminished in the middle deprivation quintile where the WIMD score tended to have a stronger effect among males (Fig 12). However, among the younger patients in the least deprived areas, males had higher risk of death than females (Fig 13).


Fig 10. Smooths from GAMs showing the association of the WIMD 2011 score on asthma death in males and females.

The shaded area represents the 95% CI of effect. CI, confidence interval; GAM, generalised additive model; WIMD, Welsh Index of Multiple Deprivation.



Fig 11. Estimated 5-year risk of asthma deaths for median age in males and females by the WIMD 2011 score.

The shaded area represents the 95% CI of risk. CI, confidence interval; Welsh Index of Multiple Deprivation.



Fig 12. Difference smooths from GAMs showing a varying gender gap in the effect of WIMD 2011 score on asthma deaths.

The shaded area represents the 95% CI of difference in effect. A positive difference means a higher effect in females than males. CI, confidence interval; GAM, generalised additive model; WIMD, Welsh Index of Multiple Deprivation.



Fig 13. Full tensor product smooths from GAMs of asthma deaths by the WIMD 2011 score and age in males and females.

The colour represents the partial effect (blue = lower, red = higher). The models did not include separate smooths for the WIMD 2011 score and age. Generally, higher deprivation and older age were associated with higher risk of asthma deaths. However, in males over the age of 75 years, lower deprivation was associated with higher risk. GAM, generalised additive model; WIMD, Welsh Index of Multiple Deprivation.


Generally, older age and higher deprivation were associated with high risk of asthma deaths. However, in males over the age of 75 years, lower deprivation was associated with higher risk (Fig 13).

Finally, deaths with asthma as the underlying cause were associated with lower income and employment levels, and there was weaker evidence that these deaths were associated with lower health and education levels. However, deaths with any mention of asthma were associated with lower scores in all those 4 domains (Fig 14).


Fig 14. GAMs of asthma death by the scores of WIMD 2011 domains, controlled for age and gender.

The shaded area represents the 95% CI of effect. CI, confidence interval; GAM, generalised additive model; WIMD, Welsh Index of Multiple Deprivation.



Interpretation of findings and comparison with other studies

A complex relationship potentially exists between the WIMD 2011 and asthma outcomes. Asthma is among limiting, long-term illnesses that contribute to the WIMD 2011 health domain [22]. The WIMD also incorporates birth outcomes, including low birth weight and preterm delivery [22]—factors linked to maternal asthma severity and medications [3739]. Asthma also potentially affects the WIMD education and employment domains as it is linked to school absenteeism [40] and job absenteeism and loss [41]. With the above direct and indirect links between asthma and the WIMD 2011, predicting asthma outcomes by the WIMD requires cautious interpretation of the findings. The socioeconomic variation in asthma-related primary care consultations and structured asthma review is marginal and may have no clinical significance. However, the wider gap in emergency asthma care and deaths indicates inequalities in disease severity and potentially in how the disease is managed. Higher asthma-related health service utilisation, especially in secondary care, often indicates more severe and/or uncontrolled disease [42].

The higher asthma-related A&E attendances in the most deprived areas might not be solely driven by more severe or worse controlled disease, but also by a complex host of factors including poorer inhaler technique and/or greater other medical/social problems (e.g., comorbidities) that worsens asthma experience. In addition, some of those A&E attendances could be driven by the tendency of people, particularly in the more deprived areas, to use A&E departments as primary care facilities. This could be possibly due to insufficient health literacy [43], which is consistent with the higher A&E attendances with lower education levels seen in our study, and/or due to GP practices and pharmacies being overcrowded or inaccessible [4446]. Accordingly, the socioeconomic contrast in asthma-related A&E attendances could overestimate the gap in asthma severity and control. Nonetheless, this gap is still evident from the clear socioeconomic gradient in asthma-related emergency admissions and deaths. However, areas with fewer local services and transport nodes, i.e., rural areas, had lower A&E attendances and admissions for asthma and slightly higher unscheduled asthma GP consultations. This may suggest that emergency care could be less accessible by those who need it who instead seek it in primary care. However, it could also be explained by the inverse correlation between the geographical access to services and the income and education levels which, as our study shows, are, in turn, inversely associated with asthma emergency care.

The 1.56 [1.18 to 2.07] risk ratio of asthma-related death between the most deprived compared to the least deprived quintiles was an average across age groups. An analysis of asthma deaths in England found similar relative risk for patients ≥45 years old (IRRs between 1.30 and 1.37) [3]. However, in that analysis, this pattern was unexpectedly reversed among the younger patients (IRR = 0.81 [0.69, 0.96]), but this was not seen in our study. Instead, the interaction we modelled between the WIMD 2011 score and age showed that among the over 75-year-old males, higher deprivation was associated with lower risk of emergency asthma admissions and deaths.

The gap in asthma severity and control could be explained by factors related to the disease, patient, and healthcare. Lack of education opportunities may have resulted in lower educational attainment and health literacy. This may hinder asthma self-management, adherence to treatment, proper inhaler technique, and engagement in clinical decision-making [1012], leading to poorer asthma control and higher dependency on healthcare, including a greater need for A&E attendances and emergency admissions [47]. The modestly higher use of both controller and reliever prescriptions with higher overall deprivation is consistent with a gradient in asthma severity and/or control. However, the corresponding inverse gradient of lower controller-to-total medication ratio—a measure of whether controllers are adequate relative to relievers—with higher deprivation suggests variations in prescribing, dosing, adherence, and/or asthma self-management [48].

Air pollution, a sub-domain in the WIMD 2011 [22], is a possible contributor to socioeconomic inequalities in asthma outcomes. A previous study in Wales found that independent measures of air pollution had weak to modest effects on “serious” asthma admissions—prolonged admissions or those followed by death from any cause [19]. While there is contradictory literature about the effect of air pollution on asthma incidence and prevalence [15,16], it was associated with higher risk of exacerbations, especially in those who live or spend time close to busy roads [13,14].

Females, especially between the age of 16 and 60 years, are generally overrepresented in primary care [49]. Nonetheless, in our study, the gender gap in asthma-related primary care consultations was minimal. However, females had higher rates of asthma-related A&E attendances, admissions in middle age, and higher risk of asthma deaths among all adults, with the inverse pattern in childhood, which is generally consistent with other studies [5054]. These patterns could be driven by gender differences in asthma development, disease experience, and outcomes [55,56].

Strength and limitations

Our study has several strengths. We used objective, real-world, person-level data with high to complete nationwide representativeness to identify most people with asthma in Wales and individually measure their asthma-related health service use. Free-of-charge healthcare, including prescriptions, limits the potential bias of patients on low income avoiding healthcare access to minimise out-of-pocket expenses [57,58]. The 2011 WIMD incorporated multifaceted deprivation domains for small areas in Wales, enabling a comprehensive assessment of socioeconomic status. We explored both the WIMD score, which allowed exploring nonlinear associations, and its quintiles. The latter approach is commonly used in the assessment of health inequalities [3,59,60]. However, it involves information loss and other methodological problems [61]. Nonetheless, in our study, both approaches led to generally consistent findings. Finally, modelling by the WIMD 2011 individual domains, each in a separate model, has provided additional insights into the drivers of asthma inequalities, which should however be interpreted in the light of collinearity between some of these domains, particularly income, employment, education, and health.

Our study has some limitations. The WIMD 2011 is an area-level index [22], and therefore, caution is required when drawing person-level inferences. However, our findings are generally consistent with other studies in the UK and elsewhere which have found similar gaps of higher incidence of asthma symptoms and emergency hospitalisations and higher asthma severity in the more deprived areas [3,6264]. Excluding patients with gaps in their primary care data from our cohort reduced the possibility of missing primary care data while insignificantly affecting the findings, as shown in the sensitivity analysis. However, this means people who died within the follow-up period, who might have more severe disease, were excluded from the main cohort in which health service utilisation was investigated. We did not exclude patients with diagnosis of chronic obstructive pulmonary disease (COPD), which may coexist with asthma, resulting in more severe symptoms and higher health service utilisation and mortality in the older ages [6567]. The AMR formula is based on the number of prescriptions which did not necessarily reflect the actual prescribed dosage (puffs per day). However, data on actual dosage are currently not available in the primary care dataset that we have used. Residual confounders and mediators might have been involved in some of the observed associations in our study. Smoking is for example associated with socioeconomic deprivation [68]. The National Survey for Wales in 2016 to 2017 has found that adults in the most deprived quintile were 3 times more likely to smoke than those in the least deprived quintile [69]. Smoking is associated with a number of limiting long-term illnesses which were accounted for in the WIMD 2011 health domain. Smoking is also associated with poor asthma outcomes. Exposure to secondhand tobacco smoke among children with asthma is associated with reduction in pulmonary function and doubles the risk of hospitalisation for asthma exacerbation [70]. Therefore, differential smoking status might have partially mediated and confounded the observed association between socioeconomic deprivation and asthma-related health service utilisation, especially in secondary care. Asthma is also associated with a range of comorbidities, such as obesity and depression, which are associated with higher asthma severity and lower control and in which a socioeconomic gradient has been observed [71,72]. However, we did not control for the potential confounding effect of comorbidities.

Implication for research, clinical practice, and public policy

The socioeconomic disparities that we have found in asthma-related health service utilisation highlight the need for multifaceted service improvement. Strategies are needed to aid optimal prescribing and prevent the excessive use of reliever inhalers [73]. In addition, the most deprived groups may require more effective health education on asthma self-management, including inhaler technique, adherence to treatment, and avoidance of triggers. However, those interventions alone are unlikely to bridge the socioeconomic gap in asthma outcomes. Rather, structural and social determinants, including the circumstances in which people are born and live, play a crucial role in asthma outcomes [74]. As with health inequalities in general, inequalities in asthma should ultimately be addressed by achieving equitable wider societal determinants of health, particularly educational opportunities, housing, and health service resourcing [75].

Avoidable health inequalities, in addition to being unfair, potentially waste resources. Given the high prevalence of asthma in Wales, even the modest gap in asthma health service utilisation, especially in secondary care, would result in avoidable, significant disease costs at the country level. To better understand and therefore tackle asthma inequalities, further research is needed to identify the most significant and modifiable determinants, estimate their avoidable financial cost to the public sector, and identify the most cost-effective service and public health interventions to reduce these inequalities and their burden.


  1. 1.
    Mukherjee M, Stoddart A, Gupta RP, Nwaru BI, Farr A, Heaven M, et al. The epidemiology, healthcare and societal burden and costs of asthma in the UK and its member nations: analyses of standalone and linked national databases. BMC Med. 2016;14(1):113. pmid:27568881
  2. 2.
    Forno E, Celedón JC. Health disparities in asthma. Am J Respir Crit Care Med. 2012;185:1033–5. pmid:22589306
  3. 3.
    Gupta RP, Mukherjee M, Sheikh A, Strachan DP. Persistent variations in national asthma mortality, hospital admissions and prevalence by socioeconomic status and region in England. Thorax. 2018;73(8):706–12. pmid:30006496
  4. 4.
    Uphoff E, Cabieses B, Pinart M, Valdés M, Antó JM, Wright J. A systematic review of socioeconomic position in relation to asthma and allergic diseases. Eur Respir J. 2014;46(2):364–74. pmid:25537562
  5. 5.
    Netuveli G, Hurwitz B, Sheikh A. Ethnic variations in incidence of asthma episodes in England & Wales: national study of 502, 482 patients in primary care. Respir Res. 2005;6(1). pmid:16242029
  6. 6.
    Hull SA, McKibben S, Homer K, Taylor SJ, Pike K, Griffiths C. Asthma prescribing, ethnicity and risk of hospital admission: an analysis of 35, 864 linked primary and secondary care records in East London. NPJ Prim Care Respir Med. 2016;26(1).
  7. 7.
    Forno E, Celedón JC. Asthma and ethnic minorities: socioeconomic status and beyond. Curr Opin Allergy Clin Immunol. 2009;9(2):154–60. pmid:19326508
  8. 8.
    Kozyrskyj AL, Kendall GE, Jacoby P, Sly PD, Zubrick SR. Association between socioeconomic status and the development of asthma: analyses of income trajectories. Am J Public Health. 2010;100(3):540–6. pmid:19696386
  9. 9.
    Cardet JC, Louisias M, King TS, Castro M, Codispoti CD, Dunn R, et al. Income is an independent risk factor for worse asthma outcomes. J Allergy Clin Immunol. 2018;141(2):754–60. pmid:28535964
  10. 10.
    Thai AL, George M. The effects of health literacy on asthma self-management. J Asthma Allergy Educ. 2010;1(2):50–5.
  11. 11.
    Mancuso CA, Rincon M. Impact of health literacy on longitudinal asthma outcomes. J Gen Intern Med. 2006;21(8):813–7. pmid:16881939
  12. 12.
    Apter AJ, Wan F, Reisine S, Bender B, Rand C, Bogen DK, et al. The association of health literacy with adherence and outcomes in moderate-severe asthma. J Allergy Clin Immunol. 2013;132(2):321–7. pmid:23591273
  13. 13.
    Orellano P, Quaranta N, Reynoso J, Balbi B, Vasquez J. Effect of outdoor air pollution on asthma exacerbations in children and adults: Systematic review and multilevel meta-analysis. PLoS ONE. 2017;12(3):1–15. pmid:28319180
  14. 14.
    Perez L, Declercq C, Iñiguez C, Aguilera I, Badaloni C, Ballester F, et al. Chronic burden of near-roadway traffic pollution in 10 European cities (APHEKOM network). Eur Respir J. 2013;42(3):594–605. pmid:23520318
  15. 15.
    Bowatte G, Lodge C, Lowe AJ, Erbas B, Perret J, Abramson MJ, et al. The influence of childhood traffic-related air pollution exposure on asthma, allergy and sensitization: a systematic review and a meta-analysis of birth cohort studies. Allergy. 2014;70(3):245–56. pmid:25495759
  16. 16.
    Gowers AM, Cullinan P, Ayres JG, Anderson HR, Strachan DP, Holgate ST, et al. Does outdoor air pollution induce new cases of asthma? Biological plausibility and evidence: a review. Respirology. 2012;17(6):887–98. pmid:22672711
  17. 17.
    Jones AP, Bentham G. Health service accessibility and deaths from asthma in 401 local authority districts in England and Wales, 1988–92. Thorax. 1997;52(3):218–22. pmid:9093335
  18. 18.
    Burr M, Verrall C, Kaur B. Social deprivation and asthma. Respir Med. 1997;91(10):603–8. pmid:9488893
  19. 19.
    Roberts SE, Button LA, Hopkin JM, Goldacre MJ, Lyons RA, Rodgers SE, et al. Influence of social deprivation and air pollutants on serious asthma. Eur Respir J. 2012;40(3):785–8. pmid:22941547
  20. 20.
    Ford DV, Jones KH, Verplancke JP, Lyons RA, John G, Brown G, et al. The SAIL Databank: building a national architecture for e-health research and evaluation. BMC Health Serv Res. 2009;9(1):157. pmid:19732426
  21. 21.
    Lyons RA, Jones KH, John G, Brooks CJ, Verplancke JP, Ford DV, et al. The SAIL databank: linking multiple health and social care datasets. BMC Med Inform Decis Mak. 2009;9:3. pmid:19149883
  22. 22.
    Welsh Assembly Government. The Welsh Index of Multiple Deprivation (WIMD Technical Report). Cardiff. 2011.
  23. 23.
    Office for National Statistics. Census geography: An overview of the various geographies used in the production of statistics collected via the UK census. [cited 2020 Feb 19]. Available from: https://www.ons.gov.uk/methodology/geography/ukgeographies/censusgeography.
  24. 24.
    British Thoracic Society and Scottish Intercollegiate Guidelines Network. British guideline on the management of asthma: A national clinical guideline. 2019.
  25. 25.
    Pearson M and Bucknall CE, editors. Measuring clinical outcome in asthma: a patient-focused approach. London: Royal College of Physicians. 1999.
  26. 26.
    NHS Wales Data Dictionary. Admission Method. Data Standards Team, NHS Wales Informatics Service. [cited 21 Mar 2020]. Available from: http://www.datadictionary.wales.nhs.uk/index.html#!WordDocuments/admissionmethod.htm.
  27. 27.
    Schatz M, Zeiger RS, Vollmer WM, Mosen D, Mendoza G, Apter AJ, et al. The controller-to-total asthma medication ratio is associated with patient-centered as well as utilization outcomes. Chest. 2006;130(1):43–50. pmid:16840381
  28. 28.
    Venables WN, Ripley BD. Modern applied statistics with S. Fourth Edition. New York: Springer, 2002.
  29. 29.
    Kleiber C, Zeileis A. Visualizing count data regressions using rootograms. Am Stat. 2016;70(3):296–303.
  30. 30.
    Wood S. mgcv: Mixed GAM computation vehicle with automatic smoothness estimation. 2020.
  31. 31.
    Simpson GL. gratia: Graceful ‘ggplot’-based graphics and other functions for GAMs fitted using ‘mgcv’. R package version 041. 2020.
  32. 32.
    von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–9. pmid:18313558
  33. 33.
    Benchimol EI, Smeeth L, Guttmann A, Harron K, Moher D, Petersen I, et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) Statement. PLoS Med. 2015;12(10):e1001885. pmid:26440803
  34. 34.
    Al Sallakh MA. Computer code for “Association of socioeconomic deprivation with asthma care, outcomes, and deaths in Wales: a five-year national linked primary and secondary care cohort study”. Available from: https://github.com/anassal/asthma-inequalities-in-wales. 2020.
  35. 35.
    Al Sallakh MA. Creating and utilising the Wales Asthma Observatory to support health policy, health service planning and clinical research. PhD thesis. Swansea University, 2018.
  36. 36.
    Al Sallakh MA, Rodgers SE, Lyons RA, Sheikh A, Davies GA. Socioeconomic deprivation and inequalities in asthma care in Wales. Lancet. 2017;390:S19.
  37. 37.
    Stenius-Aarniala BS, Hedman J, Teramo KA. Acute asthma during pregnancy. Thorax. 1996;51(4):411–4. pmid:8733495
  38. 38.
    Murphy VE, Gibson P, Talbot PI, Clifton VL. Severe asthma exacerbations during pregnancy. Obstet Gynecol. 2005;106:1046–54. pmid:16260524
  39. 39.
    Namazy JA, Murphy VE, Powell H, Gibson PG, Chambers C, Schatz M. Effects of asthma severity, exacerbations and oral corticosteroids on perinatal outcomes. Eur Respir J. 2013;41(5):1082–90. pmid:22903964
  40. 40.
    Hsu J, Qin X, Beavers SF, Mirabelli MC. Asthma-related school absenteeism, morbidity, and modifiable factors. Am J Prev Med. 2016;51(1):23–32. pmid:26873793
  41. 41.
    Hansen CL, Baelum J, Skadhauge L, Thomsen G, Omland Ø, Thilsing T, et al. Consequences of asthma on job absenteeism and job retention. Scand J Soc Med. 2012;40(4):377–84. pmid:22786923
  42. 42.
    Al Sallakh MA, Vasileiou E, Rodgers SE, Lyons RA, Sheikh A, Davies GA. Defining asthma and assessing asthma outcomes using electronic health record data: a systematic scoping review. Eur Respir J. 2017;49. pmid:28619959
  43. 43.
    Griffey RT, Kennedy SK, McGownan L, Goodman M, Kaphingst KA. Is low health literacy associated with increased emergency department utilization and recidivism? Acad Emerg Med. 2014;21(10):1109–15. pmid:25308133
  44. 44.
    Rieffe C, Oosterveld P, Wijkel D, Wiefferink C. Reasons why patients bypass their GP to visit a hospital emergency department. Accid Emerg Nurs. 1999;7:217–25. pmid:10808762
  45. 45.
    Boomla K, Hull S, Robson J. GP funding formula masks major inequalities for practices in deprived areas. BMJ. 2014;349:g7648. pmid:25515783
  46. 46.
    Baird B, Charles A, Honeyman M, Maguire D, Das P. Understanding pressures in general practice. The King’s Fund, 2016.
  47. 47.
    Bacon SL, Bouchard A, Loucks EB, Lavoie KL. Individual-level socioeconomic status is associated with worse asthma morbidity in patients with asthma. Respir Res. 2009;10(1). pmid:20017907
  48. 48.
    Federman AD, Wolf MS, Sofianou A, Martynenko M, O’Connor R, Halm EA, et al. Self-management behaviors in older adults with asthma: associations with health literacy. J Am Geriatr Soc. 2014;62(5):872–9. pmid:24779482
  49. 49.
    Wang Y, Hunt K, Nazareth I, Freemantle N, Petersen I. Do men consult less than women? An analysis of routinely collected UK general practice data. BMJ Open. 2013;3(8):e003320. pmid:23959757
  50. 50.
    Prescott E, Lange P, Vestbo J. Effect of gender on hospital admissions for asthma and prevalence of self-reported asthma: a prospective study based on a sample of the general population. Copenhagen City Heart Study Group. Thorax. 1997;52(3):287–9. pmid:9093349
  51. 51.
    Singh AK. Sex differences among adults presenting to the emergency department with acute asthma. Arch Intern Med. 1999;159(11):1237. pmid:10371232
  52. 52.
    Schatz M, Camargo CA. The relationship of sex to asthma prevalence, health care utilization, and medications in a large managed care organization. Ann Allergy Asthma Immunol. 2003;91(6):553–8. pmid:14700439
  53. 53.
    Miller MK, Lee JH, Blanc PD, Pasta DJ, Gujrathi S, Barron H, et al. TENOR risk score predicts healthcare in adults with severe or difficult-to-treat asthma. Eur Respir J. 2006;28(6):1145–55. pmid:16870656
  54. 54.
    Schatz M, Clark S, Camargo CA. Sex differences in the presentation and course of asthma hospitalizations. Chest. 2006;129(1):50–5. pmid:16424412
  55. 55.
    Zein JG, Erzurum SC. Asthma is different in women. Curr Allergy Asthma Rep. 2015;15(6). pmid:26141573
  56. 56.
    Leynaert B, Sunyer J, Garcia-Esteban R, Svanes C, Jarvis D, Cerveri I, et al. Gender differences in prevalence, diagnosis and incidence of allergic and non-allergic asthma: a population-based cohort. Thorax. 2012;67(7):625–31. pmid:22334535
  57. 57.
    Bender BG, Bender SE. Patient-identified barriers to asthma treatment adherence: responses to interviews, focus groups, and questionnaires. Immunol Allergy Clin N Am. 2005;25(1):107–30.
  58. 58.
    Soumerai SB, Ross-Degnan D, Avorn J, McLaughlin TJ, Choodnovskiy I. Effects of Medicaid drug-payment limits on admission to hospitals and nursing homes. N Engl J Med. 1991;325(15):1072–7. pmid:1891009
  59. 59.
    To T, Simatovic J, Zhu J, Feldman L, Dell SD, Lougheed MD, et al. Asthma deaths in a large provincial health system. A 10-year population-based study. Ann Am Thorac Soc. 2014;11(8):1210–7. pmid:25166217
  60. 60.
    Conrad N, Judge A, Tran J, Mohseni H, Hedgecott D, Crespillo AP, et al. Temporal trends and patterns in heart failure incidence: a population-based study of 4 million individuals. Lancet. 2018;391(10120):572–80. pmid:29174292
  61. 61.
    Harrell FE. Regression Modeling Strategies. 2020.
  62. 62.
    Watson JP, Cowen P, Lewis RA. The relationship between asthma admission rates, routes of admission, and socioeconomic deprivation. Eur Respir J. 1996;9(10):2087–93. pmid:8902471
  63. 63.
    Poyser M, Nelson H, Ehrlich R, Bateman E, Parnell S, Puterman A, et al. Socioeconomic deprivation and asthma prevalence and severity in young adolescents. Eur Respir J. 2002;19(5):892–8. pmid:12030730
  64. 64.
    Mielck A, Reitmeir P, Wjst M. Severity of childhood asthma by socioeconomic status. Int J Epidemiol. 1996;25(2):388–93. pmid:9119565
  65. 65.
    Yamauchi Y, Yasunaga H, Matsui H, Hasegawa W, Jo T, Takami K, et al. Comparison of in-hospital mortality in patients with COPD, asthma and asthma-COPD overlap exacerbations. Respirology. 2015;20(6):940–6. pmid:25998444
  66. 66.
    Bonten TN, Kasteleyn MJ, de Mutsert R, Hiemstra PS, Rosendaal FR, Chavannes NH, et al. Defining asthma–COPD overlap syndrome: a population-based study. Eur Respir J. 2017;49(5):1602008. pmid:28461292
  67. 67.
    Shantakumar S, Pwu RF, D’Silva L, Wurst K, Kuo YW, Yang YY, et al. Burden of asthma and COPD overlap (ACO) in Taiwan: a nationwide population-based study. BMC Pulm Med. 2018;18(1).
  68. 68.
    Dolman R, Gibbon R, Roberts C. Smoking in Wales: current facts. Wales Centre for Health, 2007.
  69. 69.
    Statistics for Wales. National Survey for Wales 2016–17: Population Health–Lifestyle. Welsh Government, 2017. Available from: https://gov.wales/adult-lifestyle-national-survey-walesapril-2016-march-2017.
  70. 70.
    Wang Z, May SM, Charoenlap S, Pyle R, Ott NL, Mohammed K, et al. Effects of secondhand smoke exposure on asthma morbidity and health care utilization in children: a systematic review and meta-analysis. Ann Allergy Asthma Immunol. 2015;115(5):396–401. pmid:26411971
  71. 71.
    Newton S, Braithwaite D, Akinyemiju TF. Socio-economic status over the life course and obesity: Systematic review and meta-analysis. PLoS ONE. 2017;12(5):e0177151. pmid:28510579
  72. 72.
    Lorant V. Socioeconomic inequalities in depression: a meta-analysis. Am J Epidemiol. 2003;157(2):98–112. pmid:12522017
  73. 73.
    McKibben S, Simoni AD, Bush A, Thomas M, Griffiths C. The use of electronic alerts in primary care computer systems to identify the excessive prescription of short-acting beta2-agonists for people with asthma: a systematic review. NPJ Prim Care Respir Med. 2018;28(1). pmid:29662064
  74. 74.
    Sullivan K, Thakur N. Structural and social determinants of health in asthma in developed economies: a scoping review of literature published between 2014 and 2019. Curr Allergy Asthma Rep. 2020;20(2). pmid:32030507
  75. 75.
    Health equity in England: the Marmot review 10 years on. London: Institute of Health Equity. 2020. Available from: http://www.instituteofhealthequity.org/resources-reports/marmot-review10-years-on.

Source link

Most Popular

To Top